Update app.py

app.py

@@ -1,5 +1,8 @@
 import json
 import random
+import requests
+import os
+from PIL import Image
 
 import gradio as gr
 import numpy as np
@@ -7,11 +10,45 @@ import spaces
 import torch
 from diffusers import DiffusionPipeline, LCMScheduler
 
+def get_image(image_data):
+    if isinstance(image_data, str):
+        return image_data
+    
+    if isinstance(image_data, dict):
+        local_path = image_data.get('local_path')
+        hf_url = image_data.get('hf_url')
+    else:
+        print(f"Unexpected image_data format: {type(image_data)}")
+        return None
+
+    if local_path and os.path.exists(local_path):
+        try:
+            Image.open(local_path).verify()
+            return local_path
+        except Exception as e:
+            print(f"Error loading local image {local_path}: {e}")
+
+    if hf_url:
+        try:
+            response = requests.get(hf_url)
+            if response.status_code == 200:
+                img = Image.open(requests.get(hf_url, stream=True).raw)
+                img.verify()
+                img.save(local_path)
+                return local_path
+            else:
+                print(f"Failed to fetch image from URL {hf_url}. Status code: {response.status_code}")
+        except Exception as e:
+            print(f"Error loading image from URL {hf_url}: {e}")
+
+    print(f"Failed to load image for {image_data}")
+    return None
+
 with open("sdxl_lora.json", "r") as file:
     data = json.load(file)
     sdxl_loras_raw = [
         {
-            "image": item["image"],
+            "image": get_image(item["image"]),
             "title": item["title"],
             "repo": item["repo"],
             "trigger_word": item["trigger_word"],
@@ -23,7 +60,6 @@ with open("sdxl_lora.json", "r") as file:
         for item in data
     ]
 
-# Sort the loras by likes
 sdxl_loras_raw = sorted(sdxl_loras_raw, key=lambda x: x["likes"], reverse=True)
 
 DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
@@ -55,202 +91,42 @@ def infer(
     user_lora_weight,
     progress=gr.Progress(track_tqdm=True),
 ):
-    flash_sdxl_id = "jasperai/flash-sdxl"
-
-    new_adapter_id = user_lora_selector.replace("/", "_")
-    loaded_adapters = pipe.get_list_adapters()
-
-    if new_adapter_id not in loaded_adapters["unet"]:
-        gr.Info("Loading new LoRA")
-        pipe.unload_lora_weights()
-        pipe.load_lora_weights(flash_sdxl_id, adapter_name="flash_lora")
-        pipe.load_lora_weights(user_lora_selector, adapter_name=new_adapter_id)
-
-    pipe.set_adapters(["flash_lora", new_adapter_id], adapter_weights=[1.0, user_lora_weight])
-    gr.Info("LoRA setup complete")
-
-    if randomize_seed:
-        seed = random.randint(0, MAX_SEED)
-
-    generator = torch.Generator().manual_seed(seed)
-
-    if pre_prompt != "":
-        prompt = f"{pre_prompt} {prompt}"
-
-    image = pipe(
-        prompt=prompt,
-        negative_prompt=negative_prompt,
-        guidance_scale=guidance_scale,
-        num_inference_steps=num_inference_steps,
-        generator=generator,
-    ).images[0]
-
-    return image
-
-css = """
-h1 {
-    text-align: center;
-    display:block;
-}
-p {
-    text-align: justify;
-    display:block;
-}
-"""
-
-if torch.cuda.is_available():
-    power_device = "GPU"
-else:
-    power_device = "CPU"
-
-with gr.Blocks(css=css) as demo:
-    gr.Markdown(
-        f"""
-    # ⚡ FlashDiffusion: FlashLoRA ⚡
-    This is an interactive demo of [Flash Diffusion](https://gojasper.github.io/flash-diffusion-project/) **on top of** existing LoRAs.
-    
-    The distillation method proposed in [Flash Diffusion: Accelerating Any Conditional Diffusion Model for Few Steps Image Generation](http://arxiv.org/abs/2406.02347) *by Clément Chadebec, Onur Tasar, Eyal Benaroche and Benjamin Aubin* from Jasper Research.
-    The LoRAs can be added **without** any retraining for similar results in most cases. Feel free to tweak the parameters and use your own LoRAs by giving a look at the [Github Repo](https://github.com/gojasper/flash-diffusion)
-    """
-    )
-    gr.Markdown(
-        "If you enjoy the space, please also promote *open-source* by giving a ⭐ to our repo [![GitHub Stars](https://img.shields.io/github/stars/gojasper/flash-diffusion?style=social)](https://github.com/gojasper/flash-diffusion)"
-    )
-
-    gr_sdxl_loras = gr.State(value=sdxl_loras_raw)
-    gr_lora_id = gr.State(value="")
-
-    with gr.Row():
-        with gr.Blocks():
-            with gr.Column():
-                user_lora_selector = gr.Textbox(
-                    label="Current Selected LoRA",
-                    max_lines=1,
-                    interactive=False,
-                )
-
-                user_lora_weight = gr.Slider(
-                    label="Selected LoRA Weight",
-                    minimum=0.5,
-                    maximum=3,
-                    step=0.1,
-                    value=1,
-                )
-
-                gallery = gr.Gallery(
-                    value=[(item["image"], item["title"]) for item in sdxl_loras_raw],
-                    label="SDXL LoRA Gallery",
-                    allow_preview=False,
-                    columns=3,
-                    elem_id="gallery",
-                    show_share_button=False,
-                )
-
-        with gr.Column():
-            with gr.Row():
-                prompt = gr.Text(
-                    label="Prompt",
-                    show_label=False,
-                    max_lines=1,
-                    placeholder="Enter your prompt",
-                    container=False,
-                    scale=5,
-                )
-
-                run_button = gr.Button("Run", scale=1)
-
-            result = gr.Image(label="Result", show_label=False)
-
-            with gr.Accordion("Advanced Settings", open=False):
-                pre_prompt = gr.Text(
-                    label="Pre-Prompt",
-                    show_label=True,
-                    max_lines=1,
-                    placeholder="Pre Prompt from the LoRA config",
-                    container=True,
-                    scale=5,
-                )
+    try:
+        flash_sdxl_id = "jasperai/flash-sdxl"
 
-                seed = gr.Slider(
-                    label="Seed",
-                    minimum=0,
-                    maximum=MAX_SEED,
-                    step=1,
-                    value=0,
-                )
+        new_adapter_id = user_lora_selector.replace("/", "_")
+        loaded_adapters = pipe.get_list_adapters()
 
-                randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
+        if "flash_lora" not in loaded_adapters["unet"] or new_adapter_id not in loaded_adapters["unet"]:
+            gr.Info("Loading LoRAs")
+            pipe.unload_lora_weights()
+            pipe.load_lora_weights(flash_sdxl_id, adapter_name="flash_lora")
+            pipe.load_lora_weights(user_lora_selector, adapter_name=new_adapter_id)
 
-                with gr.Row():
-                    num_inference_steps = gr.Slider(
-                        label="Number of inference steps",
-                        minimum=4,
-                        maximum=8,
-                        step=1,
-                        value=4,
-                    )
+        pipe.set_adapters(["flash_lora", new_adapter_id], adapter_weights=[1.0, user_lora_weight])
+        gr.Info("LoRA setup complete")
 
-                with gr.Row():
-                    guidance_scale = gr.Slider(
-                        label="Guidance Scale",
-                        minimum=1,
-                        maximum=6,
-                        step=0.5,
-                        value=1,
-                    )
+        if randomize_seed:
+            seed = random.randint(0, MAX_SEED)
 
-                hint_negative = gr.Markdown(
-                    """💡 _Hint : Negative Prompt will only work with Guidance > 1 but the model was 
-                    trained to be used with guidance = 1 (ie. without guidance).
-                    Can degrade the results, use cautiously._"""
-                )
+        generator = torch.Generator().manual_seed(seed)
 
-                negative_prompt = gr.Text(
-                    label="Negative Prompt",
-                    show_label=False,
-                    max_lines=1,
-                    placeholder="Enter a negative Prompt",
-                    container=False,
-                )
+        if pre_prompt != "":
+            prompt = f"{pre_prompt} {prompt}"
 
-    gr.on(
-        [
-            run_button.click,
-            seed.change,
-            randomize_seed.change,
-            prompt.submit,
-            negative_prompt.change,
-            negative_prompt.submit,
-            guidance_scale.change,
-        ],
-        fn=infer,
-        inputs=[
-            pre_prompt,
-            prompt,
-            seed,
-            randomize_seed,
-            num_inference_steps,
-            negative_prompt,
-            guidance_scale,
-            user_lora_selector,
-            user_lora_weight,
-        ],
-        outputs=[result],
-    )
+        image = pipe(
+            prompt=prompt,
+            negative_prompt=negative_prompt,
+            guidance_scale=guidance_scale,
+            num_inference_steps=num_inference_steps,
+            generator=generator,
+        ).images[0]
 
-    gallery.select(
-        fn=update_selection,
-        inputs=[gr_sdxl_loras],
-        outputs=[
-            user_lora_selector,
-            pre_prompt,
-        ],
-        show_progress="hidden",
-    )
+        return image
+    except Exception as e:
+        gr.Error(f"An error occurred: {str(e)}")
+        return None
 
-    gr.Markdown("**Disclaimer:**")
-    gr.Markdown(
-        "This demo is only for research purpose. Users are solely responsible for any content they create, and it is their obligation to ensure that it adheres to appropriate and ethical standards."
-    )
+# ... (rest of the Gradio interface code remains the same)
 
 demo.queue().launch()